Back to Models

GPT-5.3-Codex

openaiopenai/gpt-5.3-codex

API guide

Chat completion

Standard chat completions endpoint. GPT-5.3-Codex excels at open-ended agentic coding conversations.

cURL
curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-5.3-codex","messages":[{"role":"user","content":"Write a Python script to monitor system CPU and memory usage"}]}'

Tool calling

GPT-5.3-Codex supports function/tool calling for agentic workflows β€” the model can autonomously invoke tools, parse results, and decide next steps.

cURL
curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model":"openai/gpt-5.3-codex",
    "messages":[{"role":"user","content":"Check my GitHub repo for open pull requests"}],
    "tools":[{
      "type":"function",
      "function":{
        "name":"list_github_prs",
        "description":"List open pull requests for a GitHub repository",
        "parameters":{
          "type":"object",
          "properties":{
            "repo":{"type":"string","description":"Repository name"},
            "owner":{"type":"string","description":"Repository owner"}
          },
          "required":["repo","owner"]
        }
      }
    }],
    "tool_choice":"auto"
  }'

Streaming

GPT-5.3-Codex supports streaming for real-time token delivery β€” but for agentic long-running coding tasks, non-streaming often works better as the model produces larger structured outputs.

cURL
curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-5.3-codex","messages":[{"role":"user","content":"Write a Docker Compose file for a Node.js app with PostgreSQL"}],"stream":true}'

Extended reasoning

GPT-5.3-Codex supports the reasoning parameter for extended chain-of-thought. This is especially valuable for complex multi-step debugging and architecture decisions.

cURL
curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-5.3-codex","messages":[{"role":"user","content":"Debug this memory leak in the following Go code..."}],"reasoning":{"effort":"high"}}'
Fact ledger β€” every claim on this page traces here
sourceURLretrieved
Release dateopenai.com β†—2026-05-28verified
Training cutoffβ€”β€”unknown
Licenseopenai.com β†—2026-05-28verified
Speed improvement over GPT-5.2-Codexopenai.com β†—2026-05-28verified
Cybersecurity risk tieropenai.com β†—2026-05-28verified
SWE-Bench Proopenai.com β†—2026-05-28verified
Terminal-Bench 2.0llm-stats.com β†—2026-05-28to verify
OSWorld-Verifiedllm-stats.com β†—2026-05-28to verify
SWE-Lancer IC Diamondllm-stats.com β†—2026-05-28to verify
GDPvalopenai.com β†—2026-05-28verified
OpenAI introduces GPT-5.3-Codex β€” the most capable agentic coding modelopenai.com β†—2026-05-28verified
How does GPT-5.3-Codex differ from GPT-5.2-Codex?openai.com β†—2026-05-28to verify
Does GPT-5.3-Codex support vision?openai.com β†—2026-05-28to verify
Help & contact